On Oct 18, 2005, at 5:53 PM, Rob Carpenter wrote:
> Our environment consist of the following
>
> - Server 1 DL380 G4 - Xeon 3.4GHz 2GB Windows 2003 server
> Listserv HPO 14.4 and Maestro (LUI/HUB)2.0.5
>
> - Server 2 Windows 2000 server
> LSMTP 1.1B and Maestro (TRK) 2.0.5
>
> What I would like is some recommendations on what others are doing
> for a
> disaster recovery of Listserv environment similar to the one above
>
> Our Goal
>
> We would like to create a duplicate, (probably a model ProLiant
> DL380 G4 -
> Xeon 3.4GHz 3GB mem ), Listserv environment that would function as
> both a
> Test server and a Disaster recovery server for both the servers listed
> above. Listserv / LSMTP / Maestro (LUI/HUB/TRK)
>
> Is there a way to have 2 Listserv servers automatically kept in
> sync with
> each other?
> What methods are others using to create a Disaster Recovery
> environment ?
> Has anyone used PeerSync Server, High Volume Server Edition to keep
> files
> in sync.
We haven't done this for our LISTSERV server yet, but on our web and
LDAP servers, we simply replicate the data from the primary server to
a corresponding DR server in a separate building. All the software we
use for this is either licensed permanently or on a trial basis.
These are Unix systems and we use rsync to copy data to each DR
server's RAID array, which I guess is a poor man's replication
technique. We can sustain a complete failure on our web server or
main LDAP server with only a few minutes of down time while our
network services group adjusts our DNS to point everyone to the DR
server and very little data loss. I do plan to do something like this
for our LISTSERV 14.4 server (which runs with Sendmail under Red Hat
Linux), but it hasn't happened yet.
A variety of ways to replicate data are available for Windows
servers. Which method you should use depends on your organization's
budget, your network infrastructure, as well as your organization's
tolerance for down time and data loss. If you have the money
available, one of the best options is to use either a pair of iPSCSI
or fibre channel SANs (or RAIDS); one for your primary server and one
for your DR server and use the vendor's software to keep the disks in
sync.
If your employer's budget for this is modest and you can tolerate
some data loss and production down time, then recovering from nightly
tape backups will probably suffice. You just need to keep a separate
machine around (preferably some distance from your primary LISTSERV)
and keep it on parity with your primary server. Using a DR server for
test and development is not the best idea because it means that DR
server may not be ready at the last minute to actually be deployed in
a DR event. PCs are cheap so buy two: one for DR and one for R&D and
consider using the free LISTERV Lite for testing purposes and contact
L-Soft to see if a demo license for LSMTP is available.
What are you doing for backups now? If you're using an enterprise
style backup package, the vendor who sold you that package may have
some good ideas too.
By the way, the best way to avoid being harmed by a disaster (in
terms of hardware failure) is to build plenty of redundancy into as
many components in your computing infrastructure as possible. So, its
a great idea to have a DR server, but you really ought to consider
using it in production on alternate weeks, which is actually what we
do with some of our servers. The problem with DR hardware is if its
not in production, management is tempted to let it age past the point
where it can actually do duty in a DR event.
This topic is really out of the scope of this list though. If you
have any questions regarding anything I just said, feel free to write
to me directly at [log in to unmask]
|
